freebsd-dev/sys/kern/vfs_default.c

1778 lines
37 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed
* to Berkeley by John Heidemann of the UCLA Ficus project.
*
* Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bio.h>
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/event.h>
#include <sys/filio.h>
#include <sys/kernel.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/lockf.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/namei.h>
Switch the vm_object mutex to be a rwlock. This will enable in the future further optimizations where the vm_object lock will be held in read mode most of the time the page cache resident pool of pages are accessed for reading purposes. The change is mostly mechanical but few notes are reported: * The KPI changes as follow: - VM_OBJECT_LOCK() -> VM_OBJECT_WLOCK() - VM_OBJECT_TRYLOCK() -> VM_OBJECT_TRYWLOCK() - VM_OBJECT_UNLOCK() -> VM_OBJECT_WUNLOCK() - VM_OBJECT_LOCK_ASSERT(MA_OWNED) -> VM_OBJECT_ASSERT_WLOCKED() (in order to avoid visibility of implementation details) - The read-mode operations are added: VM_OBJECT_RLOCK(), VM_OBJECT_TRYRLOCK(), VM_OBJECT_RUNLOCK(), VM_OBJECT_ASSERT_RLOCKED(), VM_OBJECT_ASSERT_LOCKED() * The vm/vm_pager.h namespace pollution avoidance (forcing requiring sys/mutex.h in consumers directly to cater its inlining functions using VM_OBJECT_LOCK()) imposes that all the vm/vm_pager.h consumers now must include also sys/rwlock.h. * zfs requires a quite convoluted fix to include FreeBSD rwlocks into the compat layer because the name clash between FreeBSD and solaris versions must be avoided. At this purpose zfs redefines the vm_object locking functions directly, isolating the FreeBSD components in specific compat stubs. The KPI results heavilly broken by this commit. Thirdy part ports must be updated accordingly (I can think off-hand of VirtualBox, for example). Sponsored by: EMC / Isilon storage division Reviewed by: jeff Reviewed by: pjd (ZFS specific review) Discussed with: alc Tested by: pho
2013-03-09 02:32:23 +00:00
#include <sys/rwlock.h>
#include <sys/fcntl.h>
#include <sys/unistd.h>
#include <sys/vnode.h>
#include <sys/dirent.h>
#include <sys/poll.h>
#include <sys/stat.h>
#include <security/audit/audit.h>
#include <sys/priv.h>
#include <security/mac/mac_framework.h>
#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_page.h>
#include <vm/vm_pager.h>
#include <vm/vnode_pager.h>
2002-03-19 21:25:46 +00:00
static int vop_nolookup(struct vop_lookup_args *);
static int vop_norename(struct vop_rename_args *);
2002-03-19 21:25:46 +00:00
static int vop_nostrategy(struct vop_strategy_args *);
static int get_next_dirent(struct vnode *vp, struct dirent **dpp,
char *dirbuf, int dirbuflen, off_t *off,
char **cpos, int *len, int *eofflag,
struct thread *td);
static int dirent_exists(struct vnode *vp, const char *dirname,
struct thread *td);
#define DIRENT_MINSIZE (sizeof(struct dirent) - (MAXNAMLEN+1) + 4)
static int vop_stdis_text(struct vop_is_text_args *ap);
static int vop_stdunset_text(struct vop_unset_text_args *ap);
static int vop_stdadd_writecount(struct vop_add_writecount_args *ap);
static int vop_stdcopy_file_range(struct vop_copy_file_range_args *ap);
static int vop_stdfdatasync(struct vop_fdatasync_args *ap);
static int vop_stdgetpages_async(struct vop_getpages_async_args *ap);
static int vop_stdread_pgcache(struct vop_read_pgcache_args *ap);
static int vop_stdstat(struct vop_stat_args *ap);
static int vop_stdvput_pair(struct vop_vput_pair_args *ap);
/*
* This vnode table stores what we want to do if the filesystem doesn't
* implement a particular VOP.
*
* If there is no specific entry here, we will return EOPNOTSUPP.
*
* Note that every filesystem has to implement either vop_access
* or vop_accessx; failing to do so will result in immediate crash
* due to stack overflow, as vop_stdaccess() calls vop_stdaccessx(),
* which calls vop_stdaccess() etc.
*/
struct vop_vector default_vnodeops = {
.vop_default = NULL,
.vop_bypass = VOP_EOPNOTSUPP,
.vop_access = vop_stdaccess,
.vop_accessx = vop_stdaccessx,
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
.vop_advise = vop_stdadvise,
.vop_advlock = vop_stdadvlock,
.vop_advlockasync = vop_stdadvlockasync,
.vop_advlockpurge = vop_stdadvlockpurge,
.vop_allocate = vop_stdallocate,
.vop_deallocate = vop_stddeallocate,
.vop_bmap = vop_stdbmap,
.vop_close = VOP_NULL,
.vop_fsync = VOP_NULL,
.vop_stat = vop_stdstat,
.vop_fdatasync = vop_stdfdatasync,
.vop_getpages = vop_stdgetpages,
.vop_getpages_async = vop_stdgetpages_async,
.vop_getwritemount = vop_stdgetwritemount,
.vop_inactive = VOP_NULL,
.vop_need_inactive = vop_stdneed_inactive,
.vop_ioctl = vop_stdioctl,
.vop_kqfilter = vop_stdkqfilter,
.vop_islocked = vop_stdislocked,
.vop_lock1 = vop_stdlock,
.vop_lookup = vop_nolookup,
.vop_open = VOP_NULL,
.vop_pathconf = VOP_EINVAL,
.vop_poll = vop_nopoll,
.vop_putpages = vop_stdputpages,
.vop_readlink = VOP_EINVAL,
.vop_read_pgcache = vop_stdread_pgcache,
.vop_rename = vop_norename,
.vop_revoke = VOP_PANIC,
.vop_strategy = vop_nostrategy,
.vop_unlock = vop_stdunlock,
.vop_vptocnp = vop_stdvptocnp,
.vop_vptofh = vop_stdvptofh,
.vop_unp_bind = vop_stdunp_bind,
.vop_unp_connect = vop_stdunp_connect,
.vop_unp_detach = vop_stdunp_detach,
.vop_is_text = vop_stdis_text,
.vop_set_text = vop_stdset_text,
.vop_unset_text = vop_stdunset_text,
.vop_add_writecount = vop_stdadd_writecount,
.vop_copy_file_range = vop_stdcopy_file_range,
.vop_vput_pair = vop_stdvput_pair,
};
VFS_VOP_VECTOR_REGISTER(default_vnodeops);
/*
* Series of placeholder functions for various error returns for
* VOPs.
*/
int
vop_eopnotsupp(struct vop_generic_args *ap)
{
/*
printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name);
*/
return (EOPNOTSUPP);
}
int
vop_ebadf(struct vop_generic_args *ap)
{
return (EBADF);
}
int
vop_enotty(struct vop_generic_args *ap)
{
return (ENOTTY);
}
int
vop_einval(struct vop_generic_args *ap)
{
return (EINVAL);
}
int
vop_enoent(struct vop_generic_args *ap)
{
return (ENOENT);
}
int
vop_eagain(struct vop_generic_args *ap)
{
return (EAGAIN);
}
int
vop_null(struct vop_generic_args *ap)
{
return (0);
}
/*
* Helper function to panic on some bad VOPs in some filesystems.
*/
int
vop_panic(struct vop_generic_args *ap)
{
panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name);
}
/*
* vop_std<something> and vop_no<something> are default functions for use by
* filesystems that need the "default reasonable" implementation for a
* particular operation.
*
* The documentation for the operations they implement exists (if it exists)
* in the VOP_<SOMETHING>(9) manpage (all uppercase).
*/
/*
* Default vop for filesystems that do not support name lookup
*/
static int
vop_nolookup(ap)
struct vop_lookup_args /* {
struct vnode *a_dvp;
struct vnode **a_vpp;
struct componentname *a_cnp;
} */ *ap;
{
*ap->a_vpp = NULL;
return (ENOTDIR);
}
/*
* vop_norename:
*
* Handle unlock and reference counting for arguments of vop_rename
* for filesystems that do not implement rename operation.
*/
static int
vop_norename(struct vop_rename_args *ap)
{
vop_rename_fail(ap);
return (EOPNOTSUPP);
}
The VFS/BIO subsystem contained a number of hacks in order to optimize piecemeal, middle-of-file writes for NFS. These hacks have caused no end of trouble, especially when combined with mmap(). I've removed them. Instead, NFS will issue a read-before-write to fully instantiate the struct buf containing the write. NFS does, however, optimize piecemeal appends to files. For most common file operations, you will not notice the difference. The sole remaining fragment in the VFS/BIO system is b_dirtyoff/end, which NFS uses to avoid cache coherency issues with read-merge-write style operations. NFS also optimizes the write-covers-entire-buffer case by avoiding the read-before-write. There is quite a bit of room for further optimization in these areas. The VM system marks pages fully-valid (AKA vm_page_t->valid = VM_PAGE_BITS_ALL) in several places, most noteably in vm_fault. This is not correct operation. The vm_pager_get_pages() code is now responsible for marking VM pages all-valid. A number of VM helper routines have been added to aid in zeroing-out the invalid portions of a VM page prior to the page being marked all-valid. This operation is necessary to properly support mmap(). The zeroing occurs most often when dealing with file-EOF situations. Several bugs have been fixed in the NFS subsystem, including bits handling file and directory EOF situations and buf->b_flags consistancy issues relating to clearing B_ERROR & B_INVAL, and handling B_DONE. getblk() and allocbuf() have been rewritten. B_CACHE operation is now formally defined in comments and more straightforward in implementation. B_CACHE for VMIO buffers is based on the validity of the backing store. B_CACHE for non-VMIO buffers is based simply on whether the buffer is B_INVAL or not (B_CACHE set if B_INVAL clear, and vise-versa). biodone() is now responsible for setting B_CACHE when a successful read completes. B_CACHE is also set when a bdwrite() is initiated and when a bwrite() is initiated. VFS VOP_BWRITE routines (there are only two - nfs_bwrite() and bwrite()) are now expected to set B_CACHE. This means that bowrite() and bawrite() also set B_CACHE indirectly. There are a number of places in the code which were previously using buf->b_bufsize (which is DEV_BSIZE aligned) when they should have been using buf->b_bcount. These have been fixed. getblk() now clears B_DONE on return because the rest of the system is so bad about dealing with B_DONE. Major fixes to NFS/TCP have been made. A server-side bug could cause requests to be lost by the server due to nfs_realign() overwriting other rpc's in the same TCP mbuf chain. The server's kernel must be recompiled to get the benefit of the fixes. Submitted by: Matthew Dillon <dillon@apollo.backplane.com>
1999-05-02 23:57:16 +00:00
/*
* vop_nostrategy:
*
* Strategy routine for VFS devices that have none.
*
* BIO_ERROR and B_INVAL must be cleared prior to calling any strategy
* routine. Typically this is done for a BIO_READ strategy call.
2003-03-10 21:55:00 +00:00
* Typically B_INVAL is assumed to already be clear prior to a write
* and should not be cleared manually unless you just made the buffer
* invalid. BIO_ERROR should be cleared either way.
The VFS/BIO subsystem contained a number of hacks in order to optimize piecemeal, middle-of-file writes for NFS. These hacks have caused no end of trouble, especially when combined with mmap(). I've removed them. Instead, NFS will issue a read-before-write to fully instantiate the struct buf containing the write. NFS does, however, optimize piecemeal appends to files. For most common file operations, you will not notice the difference. The sole remaining fragment in the VFS/BIO system is b_dirtyoff/end, which NFS uses to avoid cache coherency issues with read-merge-write style operations. NFS also optimizes the write-covers-entire-buffer case by avoiding the read-before-write. There is quite a bit of room for further optimization in these areas. The VM system marks pages fully-valid (AKA vm_page_t->valid = VM_PAGE_BITS_ALL) in several places, most noteably in vm_fault. This is not correct operation. The vm_pager_get_pages() code is now responsible for marking VM pages all-valid. A number of VM helper routines have been added to aid in zeroing-out the invalid portions of a VM page prior to the page being marked all-valid. This operation is necessary to properly support mmap(). The zeroing occurs most often when dealing with file-EOF situations. Several bugs have been fixed in the NFS subsystem, including bits handling file and directory EOF situations and buf->b_flags consistancy issues relating to clearing B_ERROR & B_INVAL, and handling B_DONE. getblk() and allocbuf() have been rewritten. B_CACHE operation is now formally defined in comments and more straightforward in implementation. B_CACHE for VMIO buffers is based on the validity of the backing store. B_CACHE for non-VMIO buffers is based simply on whether the buffer is B_INVAL or not (B_CACHE set if B_INVAL clear, and vise-versa). biodone() is now responsible for setting B_CACHE when a successful read completes. B_CACHE is also set when a bdwrite() is initiated and when a bwrite() is initiated. VFS VOP_BWRITE routines (there are only two - nfs_bwrite() and bwrite()) are now expected to set B_CACHE. This means that bowrite() and bawrite() also set B_CACHE indirectly. There are a number of places in the code which were previously using buf->b_bufsize (which is DEV_BSIZE aligned) when they should have been using buf->b_bcount. These have been fixed. getblk() now clears B_DONE on return because the rest of the system is so bad about dealing with B_DONE. Major fixes to NFS/TCP have been made. A server-side bug could cause requests to be lost by the server due to nfs_realign() overwriting other rpc's in the same TCP mbuf chain. The server's kernel must be recompiled to get the benefit of the fixes. Submitted by: Matthew Dillon <dillon@apollo.backplane.com>
1999-05-02 23:57:16 +00:00
*/
static int
vop_nostrategy (struct vop_strategy_args *ap)
{
printf("No strategy for buffer at %p\n", ap->a_bp);
vn_printf(ap->a_vp, "vnode ");
ap->a_bp->b_ioflags |= BIO_ERROR;
ap->a_bp->b_error = EOPNOTSUPP;
bufdone(ap->a_bp);
return (EOPNOTSUPP);
}
static int
get_next_dirent(struct vnode *vp, struct dirent **dpp, char *dirbuf,
int dirbuflen, off_t *off, char **cpos, int *len,
int *eofflag, struct thread *td)
{
int error, reclen;
struct uio uio;
struct iovec iov;
struct dirent *dp;
KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
if (*len == 0) {
iov.iov_base = dirbuf;
iov.iov_len = dirbuflen;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = *off;
uio.uio_resid = dirbuflen;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_READ;
uio.uio_td = td;
*eofflag = 0;
#ifdef MAC
error = mac_vnode_check_readdir(td->td_ucred, vp);
if (error == 0)
#endif
error = VOP_READDIR(vp, &uio, td->td_ucred, eofflag,
NULL, NULL);
if (error)
return (error);
*off = uio.uio_offset;
*cpos = dirbuf;
*len = (dirbuflen - uio.uio_resid);
if (*len == 0)
return (ENOENT);
}
dp = (struct dirent *)(*cpos);
reclen = dp->d_reclen;
*dpp = dp;
/* check for malformed directory.. */
if (reclen < DIRENT_MINSIZE)
return (EINVAL);
*cpos += reclen;
*len -= reclen;
return (0);
}
/*
* Check if a named file exists in a given directory vnode.
*/
static int
dirent_exists(struct vnode *vp, const char *dirname, struct thread *td)
{
char *dirbuf, *cpos;
int error, eofflag, dirbuflen, len, found;
off_t off;
struct dirent *dp;
struct vattr va;
KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
found = 0;
error = VOP_GETATTR(vp, &va, td->td_ucred);
if (error)
return (found);
dirbuflen = DEV_BSIZE;
if (dirbuflen < va.va_blocksize)
dirbuflen = va.va_blocksize;
dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
off = 0;
len = 0;
do {
error = get_next_dirent(vp, &dp, dirbuf, dirbuflen, &off,
&cpos, &len, &eofflag, td);
if (error)
goto out;
if (dp->d_type != DT_WHT && dp->d_fileno != 0 &&
strcmp(dp->d_name, dirname) == 0) {
found = 1;
goto out;
}
} while (len > 0 || !eofflag);
out:
free(dirbuf, M_TEMP);
return (found);
}
int
vop_stdaccess(struct vop_access_args *ap)
{
KASSERT((ap->a_accmode & ~(VEXEC | VWRITE | VREAD | VADMIN |
VAPPEND)) == 0, ("invalid bit in accmode"));
return (VOP_ACCESSX(ap->a_vp, ap->a_accmode, ap->a_cred, ap->a_td));
}
int
vop_stdaccessx(struct vop_accessx_args *ap)
{
int error;
accmode_t accmode = ap->a_accmode;
error = vfs_unixify_accmode(&accmode);
if (error != 0)
return (error);
if (accmode == 0)
return (0);
return (VOP_ACCESS(ap->a_vp, accmode, ap->a_cred, ap->a_td));
}
/*
* Advisory record locking support
*/
int
vop_stdadvlock(struct vop_advlock_args *ap)
{
struct vnode *vp;
struct mount *mp;
struct vattr vattr;
int error;
vp = ap->a_vp;
/*
* Provide atomicity of open(O_CREAT | O_EXCL | O_EXLOCK) for
* local filesystems. See vn_open_cred() for reciprocal part.
*/
mp = vp->v_mount;
if (mp != NULL && (mp->mnt_flag & MNT_LOCAL) != 0 &&
ap->a_op == F_SETLK && (ap->a_flags & F_FIRSTOPEN) == 0) {
VI_LOCK(vp);
while ((vp->v_iflag & VI_FOPENING) != 0)
msleep(vp, VI_MTX(vp), PLOCK, "lockfo", 0);
VI_UNLOCK(vp);
}
if (ap->a_fl->l_whence == SEEK_END) {
/*
* The NFSv4 server must avoid doing a vn_lock() here, since it
* can deadlock the nfsd threads, due to a LOR. Fortunately
* the NFSv4 server always uses SEEK_SET and this code is
* only required for the SEEK_END case.
*/
vn_lock(vp, LK_SHARED | LK_RETRY);
error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
VOP_UNLOCK(vp);
if (error)
return (error);
} else
vattr.va_size = 0;
return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size));
}
int
vop_stdadvlockasync(struct vop_advlockasync_args *ap)
{
struct vnode *vp;
struct vattr vattr;
int error;
vp = ap->a_vp;
if (ap->a_fl->l_whence == SEEK_END) {
/* The size argument is only needed for SEEK_END. */
vn_lock(vp, LK_SHARED | LK_RETRY);
error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
VOP_UNLOCK(vp);
if (error)
return (error);
} else
vattr.va_size = 0;
return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size));
}
int
vop_stdadvlockpurge(struct vop_advlockpurge_args *ap)
{
struct vnode *vp;
vp = ap->a_vp;
lf_purgelocks(vp, &vp->v_lockf);
return (0);
}
/*
* vop_stdpathconf:
2003-03-10 21:55:00 +00:00
*
* Standard implementation of POSIX pathconf, to get information about limits
* for a filesystem.
* Override per filesystem for the case where the filesystem has smaller
* limits.
*/
int
vop_stdpathconf(ap)
struct vop_pathconf_args /* {
struct vnode *a_vp;
int a_name;
int *a_retval;
} */ *ap;
{
switch (ap->a_name) {
case _PC_ASYNC_IO:
*ap->a_retval = _POSIX_ASYNCHRONOUS_IO;
return (0);
case _PC_PATH_MAX:
*ap->a_retval = PATH_MAX;
return (0);
case _PC_ACL_EXTENDED:
case _PC_ACL_NFS4:
case _PC_CAP_PRESENT:
case _PC_DEALLOC_PRESENT:
case _PC_INF_PRESENT:
case _PC_MAC_PRESENT:
*ap->a_retval = 0;
return (0);
default:
return (EINVAL);
}
/* NOTREACHED */
}
/*
* Standard lock, unlock and islocked functions.
*/
int
vop_stdlock(ap)
struct vop_lock1_args /* {
struct vnode *a_vp;
int a_flags;
char *file;
int line;
} */ *ap;
2003-03-10 21:55:00 +00:00
{
struct vnode *vp = ap->a_vp;
struct mtx *ilk;
ilk = VI_MTX(vp);
return (lockmgr_lock_flags(vp->v_vnlock, ap->a_flags,
&ilk->lock_object, ap->a_file, ap->a_line));
}
/* See above. */
int
vop_stdunlock(ap)
struct vop_unlock_args /* {
struct vnode *a_vp;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
return (lockmgr_unlock(vp->v_vnlock));
}
/* See above. */
int
vop_stdislocked(ap)
struct vop_islocked_args /* {
struct vnode *a_vp;
} */ *ap;
{
return (lockstatus(ap->a_vp->v_vnlock));
}
/*
* Variants of the above set.
*
* Differences are:
* - shared locking disablement is not supported
* - v_vnlock pointer is not honored
*/
int
vop_lock(ap)
struct vop_lock1_args /* {
struct vnode *a_vp;
int a_flags;
char *file;
int line;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
int flags = ap->a_flags;
struct mtx *ilk;
MPASS(vp->v_vnlock == &vp->v_lock);
if (__predict_false((flags & ~(LK_TYPE_MASK | LK_NODDLKTREAT | LK_RETRY)) != 0))
goto other;
switch (flags & LK_TYPE_MASK) {
case LK_SHARED:
return (lockmgr_slock(&vp->v_lock, flags, ap->a_file, ap->a_line));
case LK_EXCLUSIVE:
return (lockmgr_xlock(&vp->v_lock, flags, ap->a_file, ap->a_line));
}
other:
ilk = VI_MTX(vp);
return (lockmgr_lock_flags(&vp->v_lock, flags,
&ilk->lock_object, ap->a_file, ap->a_line));
}
int
vop_unlock(ap)
struct vop_unlock_args /* {
struct vnode *a_vp;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
MPASS(vp->v_vnlock == &vp->v_lock);
return (lockmgr_unlock(&vp->v_lock));
}
int
vop_islocked(ap)
struct vop_islocked_args /* {
struct vnode *a_vp;
} */ *ap;
{
struct vnode *vp = ap->a_vp;
MPASS(vp->v_vnlock == &vp->v_lock);
return (lockstatus(&vp->v_lock));
}
/*
* Return true for select/poll.
*/
int
vop_nopoll(ap)
struct vop_poll_args /* {
struct vnode *a_vp;
int a_events;
struct ucred *a_cred;
struct thread *a_td;
} */ *ap;
{
if (ap->a_events & ~POLLSTANDARD)
return (POLLNVAL);
return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
}
/*
* Implement poll for local filesystems that support it.
*/
int
vop_stdpoll(ap)
struct vop_poll_args /* {
struct vnode *a_vp;
int a_events;
struct ucred *a_cred;
struct thread *a_td;
} */ *ap;
{
if (ap->a_events & ~POLLSTANDARD)
return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events));
return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
}
/*
* Return our mount point, as we will take charge of the writes.
*/
int
vop_stdgetwritemount(ap)
struct vop_getwritemount_args /* {
struct vnode *a_vp;
struct mount **a_mpp;
} */ *ap;
{
struct mount *mp;
struct vnode *vp;
/*
* Note that having a reference does not prevent forced unmount from
* setting ->v_mount to NULL after the lock gets released. This is of
* no consequence for typical consumers (most notably vn_start_write)
* since in this case the vnode is VIRF_DOOMED. Unmount might have
* progressed far enough that its completion is only delayed by the
* reference obtained here. The consumer only needs to concern itself
* with releasing it.
*/
vp = ap->a_vp;
mp = vfs_ref_from_vp(vp);
*(ap->a_mpp) = mp;
return (0);
}
/*
* If the file system doesn't implement VOP_BMAP, then return sensible defaults:
* - Return the vnode's bufobj instead of any underlying device's bufobj
* - Calculate the physical block number as if there were equal size
* consecutive blocks, but
* - Report no contiguous runs of blocks.
*/
int
vop_stdbmap(ap)
2003-03-10 21:55:00 +00:00
struct vop_bmap_args /* {
struct vnode *a_vp;
daddr_t a_bn;
struct bufobj **a_bop;
daddr_t *a_bnp;
int *a_runp;
int *a_runb;
} */ *ap;
{
if (ap->a_bop != NULL)
*ap->a_bop = &ap->a_vp->v_bufobj;
if (ap->a_bnp != NULL)
*ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize);
if (ap->a_runp != NULL)
*ap->a_runp = 0;
if (ap->a_runb != NULL)
*ap->a_runb = 0;
return (0);
}
int
vop_stdfsync(ap)
struct vop_fsync_args /* {
struct vnode *a_vp;
int a_waitfor;
struct thread *a_td;
} */ *ap;
{
return (vn_fsync_buf(ap->a_vp, ap->a_waitfor));
}
2003-03-10 21:55:00 +00:00
static int
vop_stdfdatasync(struct vop_fdatasync_args *ap)
{
return (VOP_FSYNC(ap->a_vp, MNT_WAIT, ap->a_td));
}
int
vop_stdfdatasync_buf(struct vop_fdatasync_args *ap)
{
return (vn_fsync_buf(ap->a_vp, MNT_WAIT));
}
/* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */
int
vop_stdgetpages(ap)
struct vop_getpages_args /* {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int *a_rbehind;
int *a_rahead;
} */ *ap;
{
return vnode_pager_generic_getpages(ap->a_vp, ap->a_m,
ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL);
}
static int
vop_stdgetpages_async(struct vop_getpages_async_args *ap)
{
int error;
error = VOP_GETPAGES(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind,
ap->a_rahead);
if (ap->a_iodone != NULL)
ap->a_iodone(ap->a_arg, ap->a_m, ap->a_count, error);
return (error);
}
int
vop_stdkqfilter(struct vop_kqfilter_args *ap)
{
return vfs_kqfilter(ap);
}
/* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */
int
vop_stdputpages(ap)
struct vop_putpages_args /* {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int a_sync;
int *a_rtvals;
} */ *ap;
{
return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
ap->a_sync, ap->a_rtvals);
}
int
vop_stdvptofh(struct vop_vptofh_args *ap)
{
return (EOPNOTSUPP);
}
int
vop_stdvptocnp(struct vop_vptocnp_args *ap)
{
struct vnode *vp = ap->a_vp;
struct vnode **dvp = ap->a_vpp;
struct ucred *cred;
char *buf = ap->a_buf;
size_t *buflen = ap->a_buflen;
char *dirbuf, *cpos;
int i, error, eofflag, dirbuflen, flags, locked, len, covered;
off_t off;
ino_t fileno;
struct vattr va;
struct nameidata nd;
struct thread *td;
struct dirent *dp;
struct vnode *mvp;
i = *buflen;
error = 0;
covered = 0;
td = curthread;
cred = td->td_ucred;
if (vp->v_type != VDIR)
return (ENOENT);
error = VOP_GETATTR(vp, &va, cred);
if (error)
return (error);
VREF(vp);
locked = VOP_ISLOCKED(vp);
VOP_UNLOCK(vp);
NDINIT_ATVP(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF, UIO_SYSSPACE,
"..", vp);
flags = FREAD;
error = vn_open_cred(&nd, &flags, 0, VN_OPEN_NOAUDIT, cred, NULL);
if (error) {
vn_lock(vp, locked | LK_RETRY);
return (error);
}
NDFREE(&nd, NDF_ONLY_PNBUF);
mvp = *dvp = nd.ni_vp;
if (vp->v_mount != (*dvp)->v_mount &&
((*dvp)->v_vflag & VV_ROOT) &&
((*dvp)->v_mount->mnt_flag & MNT_UNION)) {
*dvp = (*dvp)->v_mount->mnt_vnodecovered;
VREF(mvp);
VOP_UNLOCK(mvp);
vn_close(mvp, FREAD, cred, td);
VREF(*dvp);
vn_lock(*dvp, LK_SHARED | LK_RETRY);
covered = 1;
}
fileno = va.va_fileid;
dirbuflen = DEV_BSIZE;
if (dirbuflen < va.va_blocksize)
dirbuflen = va.va_blocksize;
dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
if ((*dvp)->v_type != VDIR) {
error = ENOENT;
goto out;
}
off = 0;
len = 0;
do {
/* call VOP_READDIR of parent */
error = get_next_dirent(*dvp, &dp, dirbuf, dirbuflen, &off,
&cpos, &len, &eofflag, td);
if (error)
goto out;
if ((dp->d_type != DT_WHT) &&
(dp->d_fileno == fileno)) {
if (covered) {
VOP_UNLOCK(*dvp);
vn_lock(mvp, LK_SHARED | LK_RETRY);
if (dirent_exists(mvp, dp->d_name, td)) {
error = ENOENT;
VOP_UNLOCK(mvp);
vn_lock(*dvp, LK_SHARED | LK_RETRY);
goto out;
}
VOP_UNLOCK(mvp);
vn_lock(*dvp, LK_SHARED | LK_RETRY);
}
i -= dp->d_namlen;
if (i < 0) {
error = ENOMEM;
goto out;
}
if (dp->d_namlen == 1 && dp->d_name[0] == '.') {
error = ENOENT;
} else {
bcopy(dp->d_name, buf + i, dp->d_namlen);
error = 0;
}
goto out;
}
} while (len > 0 || !eofflag);
error = ENOENT;
out:
free(dirbuf, M_TEMP);
if (!error) {
*buflen = i;
vref(*dvp);
}
if (covered) {
vput(*dvp);
vrele(mvp);
} else {
VOP_UNLOCK(mvp);
vn_close(mvp, FREAD, cred, td);
}
vn_lock(vp, locked | LK_RETRY);
return (error);
}
int
vop_stdallocate(struct vop_allocate_args *ap)
{
#ifdef __notyet__
struct statfs *sfs;
off_t maxfilesize = 0;
#endif
struct iovec aiov;
struct vattr vattr, *vap;
struct uio auio;
off_t fsize, len, cur, offset;
uint8_t *buf;
struct thread *td;
struct vnode *vp;
size_t iosize;
int error;
buf = NULL;
error = 0;
td = curthread;
vap = &vattr;
vp = ap->a_vp;
len = *ap->a_len;
offset = *ap->a_offset;
error = VOP_GETATTR(vp, vap, ap->a_cred);
if (error != 0)
goto out;
fsize = vap->va_size;
iosize = vap->va_blocksize;
if (iosize == 0)
iosize = BLKDEV_IOSIZE;
if (iosize > maxphys)
iosize = maxphys;
buf = malloc(iosize, M_TEMP, M_WAITOK);
#ifdef __notyet__
/*
* Check if the filesystem sets f_maxfilesize; if not use
* VOP_SETATTR to perform the check.
*/
sfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK);
error = VFS_STATFS(vp->v_mount, sfs, td);
if (error == 0)
maxfilesize = sfs->f_maxfilesize;
free(sfs, M_STATFS);
if (error != 0)
goto out;
if (maxfilesize) {
if (offset > maxfilesize || len > maxfilesize ||
offset + len > maxfilesize) {
error = EFBIG;
goto out;
}
} else
#endif
if (offset + len > vap->va_size) {
/*
* Test offset + len against the filesystem's maxfilesize.
*/
VATTR_NULL(vap);
vap->va_size = offset + len;
error = VOP_SETATTR(vp, vap, ap->a_cred);
if (error != 0)
goto out;
VATTR_NULL(vap);
vap->va_size = fsize;
error = VOP_SETATTR(vp, vap, ap->a_cred);
if (error != 0)
goto out;
}
for (;;) {
/*
* Read and write back anything below the nominal file
* size. There's currently no way outside the filesystem
* to know whether this area is sparse or not.
*/
cur = iosize;
if ((offset % iosize) != 0)
cur -= (offset % iosize);
if (cur > len)
cur = len;
if (offset < fsize) {
aiov.iov_base = buf;
aiov.iov_len = cur;
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
auio.uio_offset = offset;
auio.uio_resid = cur;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_rw = UIO_READ;
auio.uio_td = td;
error = VOP_READ(vp, &auio, ap->a_ioflag, ap->a_cred);
if (error != 0)
break;
if (auio.uio_resid > 0) {
bzero(buf + cur - auio.uio_resid,
auio.uio_resid);
}
} else {
bzero(buf, cur);
}
aiov.iov_base = buf;
aiov.iov_len = cur;
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
auio.uio_offset = offset;
auio.uio_resid = cur;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_rw = UIO_WRITE;
auio.uio_td = td;
error = VOP_WRITE(vp, &auio, ap->a_ioflag, ap->a_cred);
if (error != 0)
break;
len -= cur;
offset += cur;
if (len == 0)
break;
if (should_yield())
break;
}
out:
*ap->a_len = len;
*ap->a_offset = offset;
free(buf, M_TEMP);
return (error);
}
static int
vp_zerofill(struct vnode *vp, struct vattr *vap, off_t *offsetp, off_t *lenp,
int ioflag, struct ucred *cred)
{
int iosize;
int error = 0;
struct iovec aiov;
struct uio auio;
struct thread *td;
off_t offset, len;
iosize = vap->va_blocksize;
td = curthread;
offset = *offsetp;
len = *lenp;
if (iosize == 0)
iosize = BLKDEV_IOSIZE;
/* If va_blocksize is 512 bytes, iosize will be 4 kilobytes */
iosize = min(iosize * 8, ZERO_REGION_SIZE);
while (len > 0) {
int xfersize = iosize;
if (offset % iosize != 0)
xfersize -= offset % iosize;
if (xfersize > len)
xfersize = len;
aiov.iov_base = __DECONST(void *, zero_region);
aiov.iov_len = xfersize;
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
auio.uio_offset = offset;
auio.uio_resid = xfersize;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_rw = UIO_WRITE;
auio.uio_td = td;
error = VOP_WRITE(vp, &auio, ioflag, cred);
if (error != 0) {
len -= xfersize - auio.uio_resid;
offset += xfersize - auio.uio_resid;
break;
}
len -= xfersize;
offset += xfersize;
}
*offsetp = offset;
*lenp = len;
return (error);
}
int
vop_stddeallocate(struct vop_deallocate_args *ap)
{
struct vnode *vp;
off_t offset, len;
struct ucred *cred;
int error;
struct vattr va;
off_t noff, xfersize, rem;
vp = ap->a_vp;
offset = *ap->a_offset;
cred = ap->a_cred;
error = VOP_GETATTR(vp, &va, cred);
if (error)
return (error);
len = omin((off_t)va.va_size - offset, *ap->a_len);
while (len > 0) {
noff = offset;
error = vn_bmap_seekhole_locked(vp, FIOSEEKDATA, &noff, cred);
if (error) {
if (error != ENXIO)
/* XXX: Is it okay to fallback further? */
goto out;
/*
* No more data region to be filled
*/
offset += len;
len = 0;
error = 0;
break;
}
KASSERT(noff >= offset, ("FIOSEEKDATA going backward"));
if (noff != offset) {
xfersize = omin(noff - offset, len);
len -= xfersize;
offset += xfersize;
if (len == 0)
break;
}
error = vn_bmap_seekhole_locked(vp, FIOSEEKHOLE, &noff, cred);
if (error)
goto out;
/* Fill zeroes */
xfersize = rem = omin(noff - offset, len);
error = vp_zerofill(vp, &va, &offset, &rem, ap->a_ioflag, cred);
if (error) {
len -= xfersize - rem;
goto out;
}
len -= xfersize;
if (should_yield())
break;
}
/* Handle the case when offset is beyond EOF */
if (len < 0)
len = 0;
out:
*ap->a_offset = offset;
*ap->a_len = len;
return (error);
}
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
int
vop_stdadvise(struct vop_advise_args *ap)
{
struct vnode *vp;
struct bufobj *bo;
daddr_t startn, endn;
off_t bstart, bend, start, end;
int bsize, error;
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
vp = ap->a_vp;
switch (ap->a_advice) {
case POSIX_FADV_WILLNEED:
/*
* Do nothing for now. Filesystems should provide a
* custom method which starts an asynchronous read of
* the requested region.
*/
error = 0;
break;
case POSIX_FADV_DONTNEED:
error = 0;
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
if (VN_IS_DOOMED(vp)) {
VOP_UNLOCK(vp);
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
break;
}
/*
* Round to block boundaries (and later possibly further to
* page boundaries). Applications cannot reasonably be aware
* of the boundaries, and the rounding must be to expand at
* both extremities to cover enough. It still doesn't cover
* read-ahead. For partial blocks, this gives unnecessary
* discarding of buffers but is efficient enough since the
* pages usually remain in VMIO for some time.
*/
bsize = vp->v_bufobj.bo_bsize;
bstart = rounddown(ap->a_start, bsize);
bend = roundup(ap->a_end, bsize);
/*
* Deactivate pages in the specified range from the backing VM
* object. Pages that are resident in the buffer cache will
* remain wired until their corresponding buffers are released
* below.
*/
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
if (vp->v_object != NULL) {
start = trunc_page(bstart);
end = round_page(bend);
VM_OBJECT_RLOCK(vp->v_object);
vm_object_page_noreuse(vp->v_object, OFF_TO_IDX(start),
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
OFF_TO_IDX(end));
VM_OBJECT_RUNLOCK(vp->v_object);
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
}
bo = &vp->v_bufobj;
BO_RLOCK(bo);
startn = bstart / bsize;
endn = bend / bsize;
error = bnoreuselist(&bo->bo_clean, bo, startn, endn);
if (error == 0)
error = bnoreuselist(&bo->bo_dirty, bo, startn, endn);
BO_RUNLOCK(bo);
VOP_UNLOCK(vp);
Add the posix_fadvise(2) system call. It is somewhat similar to madvise(2) except that it operates on a file descriptor instead of a memory region. It is currently only supported on regular files. Just as with madvise(2), the advice given to posix_fadvise(2) can be divided into two types. The first type provide hints about data access patterns and are used in the file read and write routines to modify the I/O flags passed down to VOP_READ() and VOP_WRITE(). These modes are thus filesystem independent. Note that to ease implementation (and since this API is only advisory anyway), only a single non-normal range is allowed per file descriptor. The second type of hints are used to hint to the OS that data will or will not be used. These hints are implemented via a new VOP_ADVISE(). A default implementation is provided which does nothing for the WILLNEED request and attempts to move any clean pages to the cache page queue for the DONTNEED request. This latter case required two other changes. First, a new V_CLEANONLY flag was added to vinvalbuf(). This requests vinvalbuf() to only flush clean buffers for the vnode from the buffer cache and to not remove any backing pages from the vnode. This is used to ensure clean pages are not wired into the buffer cache before attempting to move them to the cache page queue. The second change adds a new vm_object_page_cache() method. This method is somewhat similar to vm_object_page_remove() except that instead of freeing each page in the specified range, it attempts to move clean pages to the cache queue if possible. To preserve the ABI of struct file, the f_cdevpriv pointer is now reused in a union to point to the currently active advice region if one is present for regular files. Reviewed by: jilles, kib, arch@ Approved by: re (kib) MFC after: 1 month
2011-11-04 04:02:50 +00:00
break;
default:
error = EINVAL;
break;
}
return (error);
}
int
vop_stdunp_bind(struct vop_unp_bind_args *ap)
{
ap->a_vp->v_unpcb = ap->a_unpcb;
return (0);
}
int
vop_stdunp_connect(struct vop_unp_connect_args *ap)
{
*ap->a_unpcb = ap->a_vp->v_unpcb;
return (0);
}
int
vop_stdunp_detach(struct vop_unp_detach_args *ap)
{
ap->a_vp->v_unpcb = NULL;
return (0);
}
static int
vop_stdis_text(struct vop_is_text_args *ap)
{
return (atomic_load_int(&ap->a_vp->v_writecount) < 0);
}
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
int
vop_stdset_text(struct vop_set_text_args *ap)
{
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
struct vnode *vp;
int n;
bool gotref;
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
vp = ap->a_vp;
n = atomic_load_int(&vp->v_writecount);
for (;;) {
if (__predict_false(n > 0)) {
return (ETXTBSY);
}
/*
* Transition point, we may need to grab a reference on the vnode.
*
* Take the ref early As a safety measure against bogus calls
* to vop_stdunset_text.
*/
if (n == 0) {
gotref = false;
if ((vn_irflag_read(vp) & VIRF_TEXT_REF) != 0) {
vref(vp);
gotref = true;
}
if (atomic_fcmpset_int(&vp->v_writecount, &n, -1)) {
return (0);
}
if (gotref) {
vunref(vp);
}
continue;
}
MPASS(n < 0);
if (atomic_fcmpset_int(&vp->v_writecount, &n, n - 1)) {
return (0);
}
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
}
__assert_unreachable();
}
static int
vop_stdunset_text(struct vop_unset_text_args *ap)
{
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
struct vnode *vp;
int n;
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
vp = ap->a_vp;
n = atomic_load_int(&vp->v_writecount);
for (;;) {
if (__predict_false(n >= 0)) {
return (EINVAL);
}
/*
* Transition point, we may need to release a reference on the vnode.
*/
if (n == -1) {
if (atomic_fcmpset_int(&vp->v_writecount, &n, 0)) {
if ((vn_irflag_read(vp) & VIRF_TEXT_REF) != 0) {
vunref(vp);
}
return (0);
}
continue;
}
MPASS(n < -1);
if (atomic_fcmpset_int(&vp->v_writecount, &n, n + 1)) {
return (0);
}
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
}
__assert_unreachable();
}
static int __always_inline
vop_stdadd_writecount_impl(struct vop_add_writecount_args *ap, bool handle_msync)
{
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
struct vnode *vp;
struct mount *mp __diagused;
int n;
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
vp = ap->a_vp;
#ifdef INVARIANTS
mp = vp->v_mount;
if (mp != NULL) {
if (handle_msync) {
VNPASS((mp->mnt_kern_flag & MNTK_NOMSYNC) == 0, vp);
} else {
VNPASS((mp->mnt_kern_flag & MNTK_NOMSYNC) != 0, vp);
}
}
#endif
n = atomic_load_int(&vp->v_writecount);
for (;;) {
if (__predict_false(n < 0)) {
return (ETXTBSY);
}
VNASSERT(n + ap->a_inc >= 0, vp,
("neg writecount increment %d + %d = %d", n, ap->a_inc,
n + ap->a_inc));
if (n == 0) {
if (handle_msync) {
vlazy(vp);
}
}
if (atomic_fcmpset_int(&vp->v_writecount, &n, n + ap->a_inc)) {
return (0);
}
Switch to use shared vnode locks for text files during image activation. kern_execve() locks text vnode exclusive to be able to set and clear VV_TEXT flag. VV_TEXT is mutually exclusive with the v_writecount > 0 condition. The change removes VV_TEXT, replacing it with the condition v_writecount <= -1, and puts v_writecount under the vnode interlock. Each text reference decrements v_writecount. To clear the text reference when the segment is unmapped, it is recorded in the vm_map_entry backed by the text file as MAP_ENTRY_VN_TEXT flag, and v_writecount is incremented on the map entry removal The operations like VOP_ADD_WRITECOUNT() and VOP_SET_TEXT() check that v_writecount does not contradict the desired change. vn_writecheck() is now racy and its use was eliminated everywhere except access. Atomic check for writeability and increment of v_writecount is performed by the VOP. vn_truncate() now increments v_writecount around VOP_SETATTR() call, lack of which is arguably a bug on its own. nullfs bypasses v_writecount to the lower vnode always, so nullfs vnode has its own v_writecount correct, and lower vnode gets all references, since object->handle is always lower vnode. On the text vnode' vm object dealloc, the v_writecount value is reset to zero, and deadfs vop_unset_text short-circuit the operation. Reclamation of lowervp always reclaims all nullfs vnodes referencing lowervp first, so no stray references are left. Reviewed by: markj, trasz Tested by: mjg, pho Sponsored by: The FreeBSD Foundation MFC after: 1 month Differential revision: https://reviews.freebsd.org/D19923
2019-05-05 11:20:43 +00:00
}
__assert_unreachable();
}
int
vop_stdadd_writecount(struct vop_add_writecount_args *ap)
{
return (vop_stdadd_writecount_impl(ap, true));
}
int
vop_stdadd_writecount_nomsync(struct vop_add_writecount_args *ap)
{
return (vop_stdadd_writecount_impl(ap, false));
}
int
vop_stdneed_inactive(struct vop_need_inactive_args *ap)
{
return (1);
}
int
vop_stdioctl(struct vop_ioctl_args *ap)
{
struct vnode *vp;
struct vattr va;
off_t *offp;
int error;
switch (ap->a_command) {
case FIOSEEKDATA:
case FIOSEEKHOLE:
vp = ap->a_vp;
error = vn_lock(vp, LK_SHARED);
if (error != 0)
return (EBADF);
if (vp->v_type == VREG)
error = VOP_GETATTR(vp, &va, ap->a_cred);
else
error = ENOTTY;
if (error == 0) {
offp = ap->a_data;
if (*offp < 0 || *offp >= va.va_size)
error = ENXIO;
else if (ap->a_command == FIOSEEKHOLE)
*offp = va.va_size;
}
VOP_UNLOCK(vp);
break;
default:
error = ENOTTY;
break;
}
return (error);
}
2003-03-10 21:55:00 +00:00
/*
* vfs default ops
* used to fill the vfs function table to get reasonable default return values.
*/
int
vfs_stdroot (mp, flags, vpp)
struct mount *mp;
int flags;
struct vnode **vpp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdstatfs (mp, sbp)
struct mount *mp;
struct statfs *sbp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdquotactl (mp, cmds, uid, arg, mp_busy)
struct mount *mp;
int cmds;
uid_t uid;
2005-12-14 00:49:52 +00:00
void *arg;
bool *mp_busy;
{
return (EOPNOTSUPP);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdsync(mp, waitfor)
struct mount *mp;
int waitfor;
{
struct vnode *vp, *mvp;
struct thread *td;
int error, lockreq, allerror = 0;
td = curthread;
lockreq = LK_EXCLUSIVE | LK_INTERLOCK;
if (waitfor != MNT_WAIT)
lockreq |= LK_NOWAIT;
/*
* Force stale buffer cache information to be flushed.
*/
loop:
MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
if (vp->v_bufobj.bo_dirty.bv_cnt == 0) {
VI_UNLOCK(vp);
continue;
}
if ((error = vget(vp, lockreq)) != 0) {
if (error == ENOENT) {
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
goto loop;
}
continue;
}
error = VOP_FSYNC(vp, waitfor, td);
if (error)
allerror = error;
vput(vp);
}
return (allerror);
}
int
vfs_stdnosync (mp, waitfor)
struct mount *mp;
int waitfor;
{
2004-07-07 07:00:02 +00:00
return (0);
}
static int
vop_stdcopy_file_range(struct vop_copy_file_range_args *ap)
{
int error;
error = vn_generic_copy_file_range(ap->a_invp, ap->a_inoffp,
ap->a_outvp, ap->a_outoffp, ap->a_lenp, ap->a_flags, ap->a_incred,
ap->a_outcred, ap->a_fsizetd);
return (error);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdvget (mp, ino, flags, vpp)
struct mount *mp;
ino_t ino;
int flags;
struct vnode **vpp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
2003-03-10 21:55:00 +00:00
int
vfs_stdfhtovp (mp, fhp, flags, vpp)
struct mount *mp;
struct fid *fhp;
int flags;
struct vnode **vpp;
{
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
2003-03-10 21:55:00 +00:00
vfs_stdinit (vfsp)
struct vfsconf *vfsp;
{
2004-07-07 07:00:02 +00:00
return (0);
}
int
vfs_stduninit (vfsp)
struct vfsconf *vfsp;
{
2004-07-07 07:00:02 +00:00
return(0);
}
int
vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace, attrname)
struct mount *mp;
int cmd;
o Change the API and ABI of the Extended Attribute kernel interfaces to introduce a new argument, "namespace", rather than relying on a first- character namespace indicator. This is in line with more recent thinking on EA interfaces on various mailing lists, including the posix1e, Linux acl-devel, and trustedbsd-discuss forums. Two namespaces are defined by default, EXTATTR_NAMESPACE_SYSTEM and EXTATTR_NAMESPACE_USER, where the primary distinction lies in the access control model: user EAs are accessible based on the normal MAC and DAC file/directory protections, and system attributes are limited to kernel-originated or appropriately privileged userland requests. o These API changes occur at several levels: the namespace argument is introduced in the extattr_{get,set}_file() system call interfaces, at the vnode operation level in the vop_{get,set}extattr() interfaces, and in the UFS extended attribute implementation. Changes are also introduced in the VFS extattrctl() interface (system call, VFS, and UFS implementation), where the arguments are modified to include a namespace field, as well as modified to advoid direct access to userspace variables from below the VFS layer (in the style of recent changes to mount by adrian@FreeBSD.org). This required some cleanup and bug fixing regarding VFS locks and the VFS interface, as a vnode pointer may now be optionally submitted to the VFS_EXTATTRCTL() call. Updated documentation for the VFS interface will be committed shortly. o In the near future, the auto-starting feature will be updated to search two sub-directories to the ".attribute" directory in appropriate file systems: "user" and "system" to locate attributes intended for those namespaces, as the single filename is no longer sufficient to indicate what namespace the attribute is intended for. Until this is committed, all attributes auto-started by UFS will be placed in the EXTATTR_NAMESPACE_SYSTEM namespace. o The default POSIX.1e attribute names for ACLs and Capabilities have been updated to no longer include the '$' in their filename. As such, if you're using these features, you'll need to rename the attribute backing files to the same names without '$' symbols in front. o Note that these changes will require changes in userland, which will be committed shortly. These include modifications to the extended attribute utilities, as well as to libutil for new namespace string conversion routines. Once the matching userland changes are committed, a buildworld is recommended to update all the necessary include files and verify that the kernel and userland environments are in sync. Note: If you do not use extended attributes (most people won't), upgrading is not imperative although since the system call API has changed, the new userland extended attribute code will no longer compile with old include files. o Couple of minor cleanups while I'm there: make more code compilation conditional on FFS_EXTATTR, which should recover a bit of space on kernels running without EA's, as well as update copyright dates. Obtained from: TrustedBSD Project
2001-03-15 02:54:29 +00:00
struct vnode *filename_vp;
int attrnamespace;
const char *attrname;
{
2004-07-07 07:00:02 +00:00
if (filename_vp != NULL)
VOP_UNLOCK(filename_vp);
2004-07-07 07:00:02 +00:00
return (EOPNOTSUPP);
}
int
vfs_stdsysctl(mp, op, req)
struct mount *mp;
fsctlop_t op;
struct sysctl_req *req;
{
return (EOPNOTSUPP);
}
static vop_bypass_t *
bp_by_off(struct vop_vector *vop, struct vop_generic_args *a)
{
return (*(vop_bypass_t **)((char *)vop + a->a_desc->vdesc_vop_offset));
}
int
vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a)
{
vop_bypass_t *bp;
int prev_stops, rc;
bp = bp_by_off(vop, a);
MPASS(bp != NULL);
prev_stops = sigdeferstop(SIGDEFERSTOP_SILENT);
rc = bp(a);
sigallowstop(prev_stops);
return (rc);
}
static int
vop_stdstat(struct vop_stat_args *a)
{
struct vattr vattr;
struct vattr *vap;
struct vnode *vp;
struct stat *sb;
int error;
u_short mode;
vp = a->a_vp;
sb = a->a_sb;
error = vop_stat_helper_pre(a);
if (error != 0)
return (error);
vap = &vattr;
/*
* Initialize defaults for new and unusual fields, so that file
* systems which don't support these fields don't need to know
* about them.
*/
vap->va_birthtime.tv_sec = -1;
vap->va_birthtime.tv_nsec = 0;
vap->va_fsid = VNOVAL;
vap->va_gen = 0;
vap->va_rdev = NODEV;
error = VOP_GETATTR(vp, vap, a->a_active_cred);
if (error)
goto out;
/*
* Zero the spare stat fields
*/
bzero(sb, sizeof *sb);
/*
* Copy from vattr table
*/
if (vap->va_fsid != VNOVAL)
sb->st_dev = vap->va_fsid;
else
sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
sb->st_ino = vap->va_fileid;
mode = vap->va_mode;
switch (vap->va_type) {
case VREG:
mode |= S_IFREG;
break;
case VDIR:
mode |= S_IFDIR;
break;
case VBLK:
mode |= S_IFBLK;
break;
case VCHR:
mode |= S_IFCHR;
break;
case VLNK:
mode |= S_IFLNK;
break;
case VSOCK:
mode |= S_IFSOCK;
break;
case VFIFO:
mode |= S_IFIFO;
break;
default:
error = EBADF;
goto out;
}
sb->st_mode = mode;
sb->st_nlink = vap->va_nlink;
sb->st_uid = vap->va_uid;
sb->st_gid = vap->va_gid;
sb->st_rdev = vap->va_rdev;
if (vap->va_size > OFF_MAX) {
error = EOVERFLOW;
goto out;
}
sb->st_size = vap->va_size;
sb->st_atim.tv_sec = vap->va_atime.tv_sec;
sb->st_atim.tv_nsec = vap->va_atime.tv_nsec;
sb->st_mtim.tv_sec = vap->va_mtime.tv_sec;
sb->st_mtim.tv_nsec = vap->va_mtime.tv_nsec;
sb->st_ctim.tv_sec = vap->va_ctime.tv_sec;
sb->st_ctim.tv_nsec = vap->va_ctime.tv_nsec;
sb->st_birthtim.tv_sec = vap->va_birthtime.tv_sec;
sb->st_birthtim.tv_nsec = vap->va_birthtime.tv_nsec;
/*
* According to www.opengroup.org, the meaning of st_blksize is
* "a filesystem-specific preferred I/O block size for this
* object. In some filesystem types, this may vary from file
* to file"
* Use minimum/default of PAGE_SIZE (e.g. for VCHR).
*/
sb->st_blksize = max(PAGE_SIZE, vap->va_blocksize);
sb->st_flags = vap->va_flags;
sb->st_blocks = vap->va_bytes / S_BLKSIZE;
sb->st_gen = vap->va_gen;
out:
return (vop_stat_helper_post(a, error));
}
static int
vop_stdread_pgcache(struct vop_read_pgcache_args *ap __unused)
{
return (EJUSTRETURN);
}
static int
vop_stdvput_pair(struct vop_vput_pair_args *ap)
{
struct vnode *dvp, *vp, **vpp;
dvp = ap->a_dvp;
vpp = ap->a_vpp;
vput(dvp);
if (vpp != NULL && ap->a_unlock_vp && (vp = *vpp) != NULL)
vput(vp);
return (0);
}